Will user based pricing make a comeback for AI? 2026 Trends
Steven Forth is a principle at Ibbaka and valueIQ. Connect on LinkedIn
Our Prediction: Over the long-term, credit based pricing will push out user based pricing.
One of the themes of the past two years is that generative AI applications and agents will finally see the demise of user based pricing as the default pricing metric.
There are a couple of reasons for this.
For most applications (outside of HRIS, Learning and Development and Personal Productivity tools) a per user or per seat price does not track value
Some AI applications propose replacing people with AIs, so that with seat based pricing the more successful the vendor is the less it would be paid
At the same time, alternative pricing models have been gaining traction and are becoming better understood. The adoption of credit based pricing by vibe coding applications like Lovable and Bolt, for example, has made credit based pricing better understood and more acceptable to buyers.
Even so, there have been a number of comments in the past month suggesting that buyers (and vendors) are frustrated with new pricing models and want to see vendors go back to per user or per seat pricing.
Salesforce CEO Mac Benioff has alluded to a return to seat-based pricing for agentic AI, after the company experimented with usage and conversation-based models.
It seems that customers were demanding for predictability and flexibility in terms of pricing, and that they have so far been fond of the company's Agentic Enterprise License Agreement (AELA).
"When we first started with Agentforce, we were talking about, oh, it's going to be so much per conversation... but customers have pushed for more flexibility," Benioff said in last week's earnings call. (See Salesforce says per-user pricing will be new AI norm)
Meanwhile, on a webinar with the head of growth for one of the vibe coding companies said something like the following …
Our buyers find credit based pricing hard to understand and they are not sure how to predict how many credits they will need. We show them how many credits will be consumed for any one action, but there is no way to know how many and what actions they will need to complete a project. This makes it harder for them to buy.
Zuora CEO Tien Tuzo shared his perspective on this in Usage Versus Seats: Has The Pricing Pendulum Swung Back?
This has led some to conclude that seat based pricing is going to make a come back and that credit based pricing and other forms of usage based pricing are overhyped.
Reading these statements, more closely, what buyers want is predictability and flexibility. User based pricing can be predictable, if you know how many users you have. It is not clear that user based pricing is particularly flexible.
Predicability and pricing
Many consider predictability to be the achilles heel of all usage based pricing models including credit based pricing. The claim is that one cannot predict usage so usage based pricing is unpredictable by nature and some variable that can be known in advance, like the number of users, is a better metric. Often left unsaid, is that many users (in an enterprise account) will make minimum use so this is a win for the vendor (and a loss for the buyer). There are workarounds, like only paying for active users or even concurrent users, but that takes us directly back to usage based pricing.
The solution is not to revert to user based pricing, which does not track value or cost, but to make usage, or at least billing for usage, predictable. How to do that.
Sell buckets of use - in most usage based, and credit based, pricing systems, the buyer commits to buying credits in advance, making pricing more predicable. Smart design of roll over and pooling rules make these fair to both buyer and seller.
Use prediction models - prediction models make it easier to guide buyers on how many credits they need or how much usage there will be. Prediction models have become easier to implement and more accurate. All agents and apps should have prediction models built in.
Connect credits to value - buyers are more willing to buy credits or pay for usage if they can make the connection to value. Price value paths rather than straight usage. A value path is a sequence of actions that results in something of value to a user.
Michael Mansard from Zuora is suggesting that all applications and agents embed or connect to their own value model. Listen to Why every SaaS app (and Agent) needs a value model
Flexibility and pricing
Buyers want predictability, but they also need flexibility. Often they do not know which users will use and app or what functionality really matters. In most cases, Pareto’s law applies: 20% of the functionality delivers 80% of the value and 80% of the use comes from 20% of the users. When we have been able to gather the data to gather the data to analyse this, which is rare, we have found that in some cases there is little connection between the users getting a lot of value and users with the most use. Value does not necessarily equal use.
These disconnects are intensified during a time of rapid innovation and change.
The standard use cases for AI enabled applications and AI agents are not set. CX (Customer Experience) and UX (User Experience) patterns have not yet settled. Read Jakob Nielsen’s substack to get one perspective on this.
Uncertainty about use cases, use patterns and value can make buyers hesitate to make large commitments, especially when that commitment is locked into some narrow definition of usage.
User based pricing is one way around this. Buy a subscription for users and let the user decide what they need. The problem with this is that buyers generally overpay (this may be why Marc Benioff like user based pricing).
Credit based pricing, where users can consume credits at their own pace and use them where they see the most value. The best design for users, and buyers, is credits that can be used across multiple agents.
In 2026 we will see this become more common, in many cases this will be delivered through wallets. Expect many vendors to introduce wallets or credit wallets in 2026 (valueIQ.ai is planning to do this).
Our prediction on user based pricing for 2026
AI driven innovation will accelerate in 2026 with more attention being paid to agent orchestration and sharing data across agents.
A few companies that are deeply committed to user based pricing, generally for legacy accounting, sales and marketing constraints, will try to go back to user based pricing.
Other companies will get more sophisticated in how they design credit models and provide predictions about usage to buyers. Some will make commitments based on those predictions.
Over the long-term, credit based pricing will push out user based pricing.
Navigating the new pricing environment brought by AI agents? Contact us @ info@ibbaka.com
